Intelligent LASSO Regression Modelling for Seaweed Drying Analysis

Seaweed, also known as macroalgae, plays a pivotal role in various industries due to its nutritional carrageenan. This study used v-Groove Hybrid Solar Drier to dry seaweed Kappaphycus Alvarezii var. Tambalang. Accurately estimating the moisture content of seaweed is essential for maximizing the drying process and product quality. The objective can be accomplished using conventional statistical and machine learning modelling methods such as Multiple Linear Regression (MLR) and LASSO regression. A multicollinearity test and stepwise regression were performed for Multiple Linear Regression to produce the best-fit MLR model. As sparse regression, LASSO regression allowed for automatic variable selection. Subsequently, the dimensionality of the dataset was decreased using Principal Component Analysis (PCA) before fitting the MLR model. This study included four key drying parameters and their interaction terms. Furthermore, the performance of these three models was evaluated and compared using the following metrics: RMSE, MAE, MSE, MAPE, R-squared and adjusted R-squared. The result of the study revealed that LASSO regression fits this seaweed drying dataset better and has predictive performance at predicting the seaweed drying moisture content. However, compared to the SVM and Random Forest regression models, LASSO would naturally perform worse than these models due to their ability to capture non-linearity in the dataset. Further investigation and development of robust and ensemble methods should be undertaken to realize the potential of LASSO regression.